solar and ultraviolet radiation - …monographs.iarc.fr/eng/monographs/vol100d/mono100d-6.pdfsolar...

35

1. Exposure Data

Terrestrial life is dependent on radiant energy from the sun. Solar radiation is largely optical radiation [radiant energy within a broad region of the electromagnetic spectrum that includes ultraviolet (UV), visible (light) and infrared radiation], although both shorter wavelength (ionizing) and longer wavelength (microwaves and radiofrequency) radiation is present. The wavelength of UV radiation (UVR) lies in the range of 100400nm, and is further subdivided into UVA (315400 nm), UVB (280315 nm), and UVC (100280 nm). The UV component of terrestrial radiation from the midday sun comprises about 95% UVA and 5% UVB; UVC and most of UVB are removed from extraterres-trial radiation by stratospheric ozone.

Approximately 5% of solar terrestrial radia-tion is UVR, and solar radiation is the major source of human exposure to UVR. Before the beginning of last century, the sun was essentially the only source of UVR, but with the advent of artificial sources the opportunity for additional exposure has increased.

1.1 Nomenclature and units

For the purpose of this Monograph, the photobiological designations of the Commission Internationale de lEclairage (CIE, International Commission on Illumination) are the most relevant, and are used throughout to define the approximate spectral regions in which certain biological absorption properties and biological interaction mechanisms may domi-nate (Commission Internationale de lEclairage, 1987).

Sources of UVR are characterized in radio-metric units. The terms dose (J/m2) and dose rate (W/m2) pertain to the energy and power, respec-tively, striking a unit surface area of an irradi-ated object (Jagger, 1985). The radiant energy delivered to a given area in a given time is also referred to as fluence, exposure dose and dose (see IARC, 1992 for further details).

A unit of effective dose [dose weighted in accordance with its capacity to bring about a particular biological effect] commonly used in cutaneous photobiology is the minimal erythemal dose (MED). One MED has been defined as the lowest radiant exposure to UVR that is sufficient to produce erythema with sharp margins 24 hours after exposure (Morison, 1983). Another end-point often used in cutaneous

SOLAR AND ULTRAVIOLET RADIATIONSolar and ultraviolet radiation were considered by a previous IARC Working Group in 1992 (IARC, 1992). Since that time, new data have become available, these have been incorpo-rated into the Monograph, and taken into consideration in the present evaluation.

IARC MONOGRAPHS 100D

photobiology is a just-perceptible reddening of exposed skin; the dose of UVR necessary to produce this minimal perceptible erythema is sometimes also referred to as a MED. In unac-climatized, white-skinned populations, there is an approximately 4-fold range in the MED of exposure to UVB radiation (Diffey & Farr, 1989). When the term MED is used as a unit of expo-sure dose, a representative value for sun-sensitive individuals of 200J/m2 is usually chosen. Since 1997, the reference action spectrum for erythema on human skin (McKinlay & Diffey, 1987) has become an International Standards Organization (ISO)/CIE norm, which, by convolution with the emission spectrum of any UVR source, enables the calculation of the erythemal yield of the source. A Standard Erythema Dose (SED) has been proposed as a unit of erythemally effective UVR dose equivalent to 100J/m2 (Commission Internationale de lEclairage, 1998).

Notwithstanding the difficulties of inter-preting accurately the magnitude of such impre-cise units as the MED and the SED, they have the advantage over radiometric units of being related to the biological consequences of the exposure.

The UV index is a tool intended for the communication of the UVR intensity to the general public. It has been developed jointly by the World Health Organization, the United Nations Environment Program, the International Commission on Non-Ionizing Radiation Protection and was standardized by ISO/CIE. It expresses the erythemal power of the sun as follows:UV Index=40 times the erythemally effective power of the sun in W/m2

The clear sky UV Index at solar noon is gener-ally in the range of 012 at the Earths surface, with values over 11 being considered extreme.

1.2 Methods for measuring UVR

UVR can be measured by chemical or physical detectors, often in conjunction with a monochro-mator or band-pass filter for wavelength selection. Physical detectors include radiometric devices, which respond to the heating effect of the radia-tion, and photoelectric devices, in which incident photons are detected by a quantum effect such as the production of electrons. Chemical detectors include photographic emulsions, actinometric solutions and UV-sensitive plastic films.

The solar UV irradiation of large portions of the Earth is currently measured using multi-frequency imaging detectors on meteorological satellites.

1.3 Sources and exposure

1.3.1 Solar UVR

Optical radiation from the sun is modified substantially as it passes through the Earths atmosphere, although about two-thirds of the energy from the sun that enters the atmosphere penetrates to ground level. The annual variation in extraterrestrial radiation is less than 10%; the variation in the modifying effect of the atmos-phere is far greater (Moseley, 1988).

On its path through the atmosphere, solar UVR is absorbed and scattered by various constituents of the atmosphere. It is scattered by air molecules, particularly oxygen and nitrogen, by aerosol and dust particles, and is scattered and absorbed by atmospheric pollution. Total solar irradiance and the relative contributions of different wavelengths vary with altitude. Clouds attenuate solar radiation, although their effect on infrared radiation is greater than on UVR. Reflection of sunlight from certain ground surfaces may contribute significantly to the total amount of scattered UVR (Moseley, 1988).

The levels of solar UVB radiation reaching the surface of the Earth are largely controlled

36

Solar and UV radiation

by the stratospheric ozone layer, which has been progressively depleted as a result of accumula-tion of ozone-destroying chemicals in the Earths atmosphere mostly chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs), whose main use has been in refrigeration and air-conditioning. The accumulation of ozone-depleting chemicals in the atmosphere ceased largely as a result of the Montreal Protocol on Substances that deplete the ozone layer, which was opened for signature in 1987, and has been ratified by 196 states.

Global climate change due to the accumula-tion of carbon dioxide (CO2) in the atmosphere can also adversely affect stratospheric ozone. This will influence whether, when, and to what extent ozone levels will return to pre-1980 values. The current best estimate is that global (60S60N) ozone levels will return to pre-1980 levels around the middle of the 21st century, at or before the time when stratospheric concentrations of ozone-depleting gases return to pre-1980 levels. Climate change will also influence surface UV radiation through changes induced mainly to clouds and the ability of the Earths surface to reflect light. Aerosols and air pollutants are also expected to change in the future. These factors may result in either increases or decreases of surface UV irradiance, through absorption or scattering. As ozone depletion becomes smaller, these factors are likely to dominate future UV radiation levels (World Meteorological Organization, 2007).

The amount of solar UVR measured at the Earths surface depends upon several factors as follows:

Time of day: In summer, about 2030% of the total daily amount of UVR is received between 11:00 and 13:00, and 75% between 9:00 and 15:00 (sun time not local time; Diffey, 1991).

Season: Seasonal variation in terrestrial UV irradiance, especially UVB, at the Earths surface is significant in temperate

regions but much less nearer the equator (Diffey, 1991).

Geographic latitude: Annual UVR expo-sure dose decreases with increasing dis-tance from the equator (Diffey, 1991).

Altitude: In general, each 300 metre increase in altitude increases the sun-burning effectiveness of sunlight by about 4% (Diffey, 1990).

Clouds: Clouds influence UV ground irradiance, through reflection, refrac-tion, absorption and scattering, and may increase or, more usually, decrease UV ground irradiance. Complete light cloud cover prevents about 50% of UVR energy from reaching the surface of the Earth (Diffey, 1991). Very heavy cloud cover absorbs and can virtually eliminate UVR even in summer. Even with heavy cloud cover, however, the scattered UVR component of sunlight (as opposed to that coming directly from the sun) is seldom less than 10% of that under clear sky. While most clouds block some UV radia-tion, the degree of protection depends on the type and amount of clouds; some clouds can actually increase the UV intensity on the ground by reflecting, refracting and scattering the suns rays. For example, under some circumstances (haze, cirrus skies, solar zenith angles ranging from 4063), the solar irradi-ance at Toowoomba, Australia (27.6S, 151.9E), was found to be 8% greater than that of an equivalent clear sky (Sabburg & Wong, 2000; Sabburg et al., 2001).

Surface reflection: The contribution of reflected UVR to a persons total UVR exposure varies in importance with sev-eral factors. A grass lawn scatters 25% of incident UVB radiation. Sand reflects about 1015%, so that sitting under an umbrella on the beach can lead to sun-burn both from scattered UVB from the

37


sky and reflected UVB from the sand. Fresh snow may reflect up to 8590% of incident UVB radiation while water, in particular white foam in the sea, may reflect up to 30%. Ground reflectance is important, because parts of the body that are normally shaded are exposed to reflected radiation (Diffey, 1990).

Air pollution: Tropospheric ozone and other pollutants can decrease UVR.

(a) Measurements of terrestrial solar radiation

Because UVR wavelengths between about 295320 nm (UVB radiation) in the terrestrial solar spectrum are thought to be those mainly responsible for adverse health effects, several studies have focused on this spectral region. Accurate measurements of UVR in this spectral band are difficult to obtain, however, because the spectral curve of terrestrial solar irradiance increases by a factor of more than five between 290320 nm. Nevertheless, extensive measure-ments of ambient UVR in this spectral band have been made worldwide. Measurements of terres-trial solar UVA are less subject to error than measurements of UVB, because the spectrum does not vary widely with zenith angle and the spectral irradiance curve is relatively flat (IARC, 1992).

The total solar radiation that arrives at the Earths surface is termed global radiation. Global radiation is made up of two components, referred to as direct and diffuse. Approximately 70% of the UVR at 300nm is in the diffuse component rather than in the direct rays of the sun. The ratio of diffuse to direct radiation increases steadily from less than 1.0 at 340 nm to at least 2.0 at 300 nm. UVR reflected from the ground (the albedo) may also be important (IARC, 1992).

Solar UV levels reaching the Earths surface can now be measured by satellites using hyper-spectral imaging to observe solar backscatter

radiation in the visible and ultraviolet ranges. NASAs Total Ozone Mapping Spectrometer (TOMS) device was installed on several space-craft, including the Earth Probe spacecraft for collecting data during 19962005. TOMS is no longer available but the continuity of satellite-derived global UV data is maintained via the new Ozone Monitoring Instrument (OMI), on board the Aura satellite (http://aura.gsfc.nasa.gov/index.html). The presence of aerosols, clouds and snow or ice cover can lead to significant biases, and new algorithms have been developed to improve the satellite-derived measurement of surface UV irradiance using Advanced Very High Resolution Radiometer (AVHRR) and Meteosat images. Currently the European Solar Data Base (SoDa) is capable to perform on-the-fly fast inter-polation with a non-regular grid and to provide data for any geographic site with a limitation to a 5-km grid cell. The SoDa contains information going back to the year 1985, available at http://www.soda-is.com/eng/services/services_radia-tion_free_eng.php.

Satellite data have been used to draw maps of UV exposure, and are available for use for epide-miological and other purposes. For example, data sets of UV irradiance derived from TOMS data for the period 1979 to 2000 are available by date, latitude and longitude for UVB and UVA. Data from satellites and ground-level measurements show that UV irradiation does not vary steadily with latitude but that local conditions may greatly influence actual UV irradiation levels (a good example of this situation may be found in the extremely elevated UV levels recorded in the summer 2003 during the heat wave that killed thousands of people in France and Northern Italy).

(b) Personal exposures

Individual sun exposure can be estimated through questionnaires, which are at best semi-quantitative, and do not give any detailed

38

http://aura.gsfc.nasa.gov/index.htmlhttp://aura.gsfc.nasa.gov/index.htmlhttp://www.soda-is.com/eng/services/services_radiation_free_eng.phphttp://www.soda-is.com/eng/services/services_radiation_free_eng.phphttp://www.soda-is.com/eng/services/services_radiation_free_eng.php


information on the wavelength of UV exposure. Individual UV dosimeters have been used in epidemiological studies, but cannot be used for the large-scale monitoring of UV exposure of populations.

Exposure data for different anatomical sites is of value in developing biological doseresponse relationships. The exposure of different anatomical sites to solar UVR depends not only on ambient UVR and the orientation of sites with respect to the sun, but also on cultural and social behaviour, type of clothing, and use of sunscreen. The most exposed skin surfaces, such as the nose, tops of the ears and forehead, have levels of UVB exposure that range up to one order of magnitude relative to that of the lesser exposed areas, such as underneath the chin. Ground reflectance plays a major role in expo-sure to UVB of all exposed body parts, including the eye and shaded skin surfaces, particularly with highly reflective surfaces such as snow. The solar exposure of the different anatomical sites of outdoor workers has recently been calculated (Milon et al., 2007) [Computerised models that integrate direct, diffuse and reflected radiation are currently being developed].

Sunscreens can be applied to control the dose of UVR to exposed skin. While undoubtedly useful when sun exposure is unavoidable (IARC, 2001), their use may lead to a longer duration of sun exposure when sun exposure is intentional (Autier et al., 2007).

The cumulative annual exposure dose of solar UVR varies widely among individuals in a given population, depending to a large extent on the occupation and extent of outdoor activities. For example, it has been estimated that indoor workers in mid-latitudes (4060N) receive an annual exposure dose of solar UVR to the face of about 40160 times the MED, depending on their level of outdoor activities, whereas the annual solar exposure dose for outdoor workers is typically around 250 times the MED. Because few actual measurements of personal exposures

have been reported, these estimates should be considered to be very approximate. They are also subject to differences in cultural and social behaviour, clothing, occupation, and outdoor activities.

1.3.2 Artificial sources of UVR

Cumulative annual outdoor exposure may be increased by exposure to artificial sources of UVR. Indoor tanning is a widespread practice in most developed countries, particularly in northern Europe and the United States of America, and is gaining popularity even in sunny countries like Australia. The prevalence of indoor tanning varies greatly among different countries, and has increased during the last decades (IARC, 2006a). The majority of users are young women, and a recent survey indicated that in the USA, up to 11% of adolescents aged 11years had ever used an indoor tanning device (Cokkinides et al., 2009). The median annual exposure dose from artifi-cial tanning is probably 2030 times the MED. Prior to the 1980s, tanning lamps emitted high proportions of UVB and even UVC. Currently used appliances emit primarily UVA; and in countries where tanning appliances are regu-lated (e.g. Sweden and France), there is a 1.5% upper limit UVB. However, commercially avail-able natural UV-tanning lamps may emit up to 4% UVB. UV emission of a modern tanning appliance corresponds to an UV index of 12, i.e. equivalent to midday tropical sun (IARC, 2006a).

Other sources of exposures to UVR include medical and dental applications. UVR has been used for several decades to treat skin diseases, notably psoriasis. A variety of sources of UVR are used, emitting either broad-band UVA or narrow-band UVB. A typical dose in a single course of UVB phototherapy can be in the range of 200300 times the MED (IARC, 2006a).

UVR is also used in many different indus-tries, yet there is a paucity of data concerning human exposure from these applications,

39


probably because in normal practice, sources are well contained and exposure doses are expected to be low. In some settings, workers may be exposed to radiation by reflection or scattering from adjacent surfaces. Staff in hospitals who work with unenclosed phototherapy equipment are at potential risk of overexposure unless protective measures are taken. Indoor tanning facilities may comprise 20 or more UVA tanning appliances, thus potentially exposing operators to high levels (>20W/m2) of UVA (IARC, 2006a).

Acute overexposures to the eyes are common among electric arc welders. Individuals exposed to lighting from fluorescent lamps may typi-cally receive annual exposure doses of UVR in the range of 030 times the MED, depending on illuminance levels and whether or not the lamps are housed behind plastic diffusers. It is also worth noting that tungstenhalogen lamps used for general lighting may emit broad-band UVR (including UVC) when not housed behind a glass filter.

2. Cancer in Humans

2.1 Natural sunlight

2.1.1 Basal cell carcinoma and cutaneous squamous cell carcinoma

In the previous IARC Monograph (IARC, 1992), the evaluation of the causal association of basal cell carcinoma and squamous cell carci-noma with solar radiation was based on descrip-tive data in Caucasian populations, which showed positive associations with birth and/or residence at low latitudes and rare occurrence at non-sun-exposed anatomical sites. The evaluation was also based on casecontrol and cohort studies whose main measures were participants retro-spectively recalled sun exposure. The majority of analyticalal studies published since have also used recalled amount of sun exposure, though

some more recent studies have made objective measures of ambient UV and used clinical signs of cumulative UV damage to the skin such as solar lentigines and actinic keratoses (Table2.1 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.1.pdf, Table 2.2 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.2.pdf, and Table 2.3 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.3.pdf).

With regard to basal cell carcinoma, all studies except one (Corona et al., 2001) showed significant positive associations with sunburns at some stage of life or overall. Of the studies that collected information on the presence of actinic keratoses (Green et al., 1996; Corona et al., 2001; Walther et al., 2004; Pelucchi et al., 2007), all showed this also to be a strong risk factor (Tables2.1 and 2.3 on-line). It was proposed that the association of basal cell carcinoma with sun exposure may vary by histological subtype and anatomical site (Bastiaens et al., 1998). Although a casecontrol study showed this variation for recalled sun exposure (Pelucchi et al., 2007), a cohort study did not (Neale et al., 2007).

For squamous cell carcinoma, while casecontrol studies tended to demonstrate little asso-ciation with sunburns (Table2.2 on-line), cohort studies uniformly showed significant positive associations (Table 2.3 on-line). The presence of actinic keratoses, a proportion of which are squamous cell carcinoma precursors, was the strongest risk factor identified (Table2.3 on-line; Green et al., 1996).

2.1.2 Cutaneous malignant melanoma

Cutaneous malignant melanoma occurs in the pigment cells of the skin. Until 1015 years ago, with the exception of two histological subgroups, melanoma was usually regarded as a single entity in analytical studies assessing the association with sunlight. The two subgroups,

40

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.1.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.1.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.2.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.2.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.2.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.3.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.3.pdf


lentigo maligna melanoma and acral lentiginous melanoma, were usually excluded from studies, the former paradoxically because of its known causal link with cumulative sun exposure, the latter for the opposite reason because it typically occurs on the soles of the feet.

In the previous IARC Monograph (IARC, 1992), the evaluation of the causal association between solar radiation and melanoma was based on descriptive data and on data from casecontrol studies. The main measures of exposure were participants recalled sun expo-sure. Intermittent sun exposure, which loosely equated with certain sun-intensive activities, such as sunbathing, outdoor recreations, and holidays in sunny climates, generally showed moderate-to-strong positive associations with melanoma. However, chronic or more contin-uous exposure, which generally equated with occupational exposure, and total sun expo-sure (sum of intermittent+chronic), generally showed weak, null or negative associations.

These results were collectively interpreted under the intermittent sun exposure hypothesis (Fears et al., 1977) as showing that melanoma occurs as a result of a pattern of intermittent intense sun exposure rather than of more contin-uous sun exposure. Studies that had also assessed objective cutaneous signs of skin damage that were generally assumed to be due to accumulated sun exposure, e.g. presence or history of actinic keratoses, or signs of other sun-related skin damage, showed, almost uniformly, strong posi-tive associations with melanoma. This inconsist-ency of evidence with the apparently negative associations of reported chronic sun exposure with melanoma was noted but not satisfactorily explained.

Several systematic reviews and meta-anal-yses of analytical studies of the association of melanoma with sun exposure have been published since (Table2.4 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.4.pdf). The summary melanoma relative

risk (RR) estimates of one of the largest meta-analyses, based on 57 studies published up to September 2002 (Gandini et al.,, 2005a, b) were: sunburn (ever/never), 2.0 (95%CI: 1.72.4); inter-mittent sun exposure (high/low), 1.6 (95%CI: 1.32.0); chronic sun exposure (high/low), 1.0 (95%CI: 0.91.0); total sun exposure (high/low), 1.3 (95%CI: 1.01.8); actinic tumours (present, past/none), 4.3 (95%CI: 2.86.6).

Casecontrol studies and the cohort study (Veierd et al., 2003) that have been published since September 2002 have shown results that are generally consistent with the meta-anal-ysis, and have not been included in this review (Table 2.5 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.5.pdf and Table 2.6 available at http://monog r aphs . ia rc . f r/ ENG/Monog r aphs/vol100D/100D-01-Table2.6.pdf).

(a) Anatomical site of melanoma

Melanomasun-exposure associations according to the anatomical site of the melanoma have recently gained greater consideration. Several studies reported differences in age-specific incidence rates by site of melanoma (Holman et al., 1980; Houghton et al., 1980; Elwood & Gallagher, 1998; Bulliard & Cox, 2000). The numerous analytical studies of risk factors by site of melanoma (Weinstock et al., 1989; Urso et al., 1991; Green, 1992; Krger et al., 1992; Rieger et al., 1995; Whiteman et al., 1998; Carli et al., 1999; Hkansson et al., 2001; Winnepenninckx & van den Oord, 2004; Cho et al., 2005; Purdue et al., 2005; Nikolaou et al., 2008) collectively show that melanomas of the head and neck are strongly associated with actinic keratoses, and melanomas on the trunk are strongly associ-ated with naevi. Similar findings have been reported from recent detailed casecase studies (Whiteman et al., 2003, 2006; Siskind et al., 2005; Lee et al., 2006).

41

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.4.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.4.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.4.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.5.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.5.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.5.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.6.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.6.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.6.pdf


(b) Skin pigmentation

Two observations from epidemiological studies may help explain the paradox of the lack of association of melanoma with chronic sun exposure. First, outdoor workers are not at a substantially increased risk of melanoma (IARC, 1992; Armstrong & Kricker, 2001); second, outdoor workers tend to have a higher-than-average ability to develop a tan (Green et al., 1996; Chang et al., 2009). Outdoor workers tend to be constitutionally protected from solar skin damage and at a lower risk of skin cancer than workers in other occupations because of self-selection based on skin pigmentation. Indeed, such self-selection has been observed in a non-Hispanic white study population from Philadelphia and San Francisco, USA, whereby the average number of hours outdoors in general increases with an increasing ability to tan (Fears et al., 2002). The role of baseline sun sensitivity in influencing sun exposure in the etiology of melanoma has long been recognized (Holman et al., 1986; Nelemans et al., 1995).

(c) Latitude

The assessment and reporting of sun expo-sure may vary among studies at different lati-tudes, due to latitude differences in sun exposure opportunity and behaviour (Elwood & Diffey, 1993; Gandini et al., 2005a, b). One approach to avoid the problems of quantifying individual sun exposure at different latitudes has been to use ambient UV flux (Fears et al., 2002; Kricker et al., 2007) for individuals through life, calculated from their residential histories, to accurately quantify at least potential solar UV exposure.

Two casecontrol studies, both done at comparatively high latitudes (Connecticut, USA; Chen et al., 1996) and (Italy; Naldi et al., 2005), and one pooled analysis stratified by latitude (Chang et al., 2009), have presented site-specific melanoma risk estimates in relation to latitude (see Table 2.5 on-line). Recalls of sunburns

throughout life were generally predictive of melanomas at all sites in both casecontrol studies and in the pooled analysis (RR, 1.02.0). Those who had objective signs of cumulative sun damage were at increased risk of melanoma at specific sites: the presence of solar lentigines increased the risk of melanoma on the lower limbs (Naldi et al., 2005; RR, 1.5; 95%CI: 1.02.1, with reference to absence of solar lentigines), while actinic keratoses increased the risk of melanoma on the head and neck (Chang et al., 2009; RR, 3.1; 95%CI: 1.46.7; based on three studies from high to low latitudes in which solar keratoses were measured). [The Working Group noted that the omission from many studies of the lentigo maligna melanoma subgroup, which is known to be associated with cumulative sun exposure, potentially results in an underestima-tion of the association with melanomas on the head and limbs.]

2.1.3 Cancer of the lip

Cancer of the lip has been associated with outdoor occupations in several descriptive studies (IARC, 1992). Three early casecontrol studies reported increases in risk for cancer of the lip with outdoor work, but use of tobacco could not be ruled out as an explanation for this association in any study (Keller, 1970; Spitzer et al., 1975; Dardanoni et al., 1984).

Two casecontrol studies have been published since that include information on tobacco smoking. The first (Pogoda & Preston-Martin, 1996), which included women only, found increased risks of cancer of the lip with average annual residential UV flux, recalled average annual hours spent in outdoor activities, and having played high-school or college sports; risk estimates were adjusted for complexion, history of skin cancer and average number of cigarettes smoked per day. Risk was not increased in women whose last occupation was outdoors (odds ratio (OR)), 1.2; 95%CI: 0.52.8). The doseresponse

42


relationship with recalled average annual hours spent in outdoor activities was inconsistent: with 300 hours. The second, which included men only (Perea-Milla Lpez et al., 2003), found no evidence of an increased risk for cancer of the lip with estimates of cumulative sun exposure during leisure time or holiday. Risk was increased with cumulative sun exposure in outdoor work during the summer months, but without any doseresponse (OR, 11.712.7; with wide confidence intervals). The odds ratios were adjusted for cumulative alcohol and tobacco intake and leaving the cigarette on the lip, among other things. In a meta-analysis of cancer in farmers (Acquavella et al., 1998), the pooled relative risk for cancer of the lip from 14 studies was 1.95 (95%CI: 1.822.09) (P for heter-ogeneity among studies, 0.22). [The Working Group noted that given the relative risks for oesophageal cancer and lung cancer were 0.77 and 0.65, respectively, confounding by smoking was unlikely, but confounding with other farm-related exposures could not be excluded.]

See Table2.7 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.7.pdf and Table 2.8 available at http://monog r aphs . ia rc . f r/ ENG/Monog r aphs/vol100D/100D-01-Table2.8.pdf.

2.1.4 Cancer of the eye

(a) Squamous cell carcinoma of the conjunctiva

(i) Descriptive studiesIncidence of squamous cell carcinoma of the

eye was inversely correlated with latitude across a wide range of countries (Newton et al., 1996), and directly associated with measured ambient UVB irradiance across the original nine Surveillance Epidemiology and End Results (SEER) cancer registry areas of the USA (Sun et al., 1997).

(ii) Casecontrol studiesThree small casecontrol studies included

only or mainly cases with conjunctival intraepi-thelial neoplasia (Table 2.9 available at http://monog r aphs . ia rc . f r/ ENG/Monog r aphs/vol100D/100D-01-Table2.9.pdf). Napora et al. (1990) compared 19 patients with biopsy-proven conjunctival intraepithelial neoplasia (including one with invasive squamous cell carcinoma) with 19 age- and sex-matched controls. The odds ratio for office work was 0.21 [95%CI: 0.040.99; Fisher Exact 95%CIs calculated from numbers in authors table]. Lee et al. (1994) included 60 [probably prevalent] cases of ocular surface epithelial dysplasia (13 were conjunctival squamous cell carcinoma) diagnosed over 19 years (40% participation), and 60 age- and sex-matched hospital-based controls. Among others, positive associations were observed between ocular surface epithelial dysplasia and history of solar keratoses [OR, for history at 50% of daytime was spent outdoors were similarly but more weakly associated with ocular surface epithelial dysplasia. Tulvatana et al. (2003) studied 30 cases of conjunctival squa-mous cell neoplasia (intraepithelial or invasive) and 30 age- and sex-matched control patients having extracapsular cataract extraction from whom diseased conjunctiva was taken [site of biopsy not specified]. Solar elastosis [repre-senting pathologically proven solar damage] was observed in the conjunctiva of 53% of cases and 3% of controls, resulting in an odds ratio of 16.0 (95%CI: 2.49671). [The Working Group noted that while pathologists were said to be masked, it was not stated that tissue sections from cases were free of neoplastic tissue.]

43

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.7.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.7.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.7.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.8.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.8.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.8.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.9.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.9.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.9.pdf


In the only casecontrol study of exclusively conjunctival squamous cell carcinoma, Newton et al. (2002) studied 60 Ugandan patients with a clinical diagnosis of conjunctival squamous cell carcinoma and 1214 controls diagnosed with other cancers not known to be associated with solar UV exposure or infection with HIV, HPV or Kaposi Sarcoma herpesvirus. The risk for conjunctival squamous cell carcinoma increased with time spent cultivating: with reference to 09 hours a week, the odds ratios were 1.9 for 1019 hours and 2.4 for 20 hours (P=0.05), adjusted for age, sex, region of residence, HIV-1 status, and low personal income. Both HIV-1 status and personal income were strong predic-tors of risk.

(b) Ocular melanoma

(i) Descriptive studiesNo increase in the incidence of ocular

melanoma was recorded by the US SEER programme during 197498, which is in contrast with the increasing incidence of cutaneous melanoma over the same period (Inskip et al., 2003).

Three studies have reported on the distri-bution of choroidal melanomas within the eye in relation to the presumed distribution of choroidal sun exposures across the choroid. The first of these (Horn et al., 1994), which analysed 414 choroidal, 20 ciliary body and 18 iris melanomas, concluded that choroidal and iris melanomas were located most frequently in the areas that are presumably exposed to the most sunlight. Specifically, melanomas in the posterior choroid were observed to preferentially involve the central area. The second (Schwartz et al., 1997), which analysed 92 choroidal mela-nomas, concluded that there was no preferential location for tumours on the choroid, having rigorously estimated the average dose distribu-tion on the retina received in outdoor daylight. A third study (Li et al., 2000), which analysed 420

choroidal and ciliary body melanomas, mapped incident melanomas on the retina and observed that rates of occurrence were concentrated in the macula area, and decreased progressively with increasing distance from the macula to the ciliary body. It was concluded that this pattern was consistent with the dose distribution of light on the retinal sphere as estimated by Schwartz et al. (1997).

(ii) Casecontrol and cohort studiesNine casecontrol studies and one cohort

study reported on associations of sun exposure with ocular melanoma (Gallagher et al., 1985; Tucker et al., 1985; Holly et al., 1990; Seddon et al., 1990; van Hees et al., 1994; Pane & Hirst, 2000; Hkansson et al., 2001; Vajdic et al., 2002; Lutz et al., 2005 (incorporating also data from Gunel et al., 2001); and Schmidt-Pokrzywniak et al., 2009). In addition, one previously reported casecontrol study reported new analyses of occupation and ocular melanoma (Holly et al., 1996; Tables2.8 and 2.9 on-line).

Four studies (Gallagher et al., 1985; Holly et al., 1990; Seddon et al., 1990; Tucker et al., 1985) found an increased risk for ocular melanoma in people with light skin, light eye colour or light hair colour. Outdoor activities were associated with ocular melanoma in one study (Tucker et al., 1985).

Four studies (Tucker et al., 1985; Seddon et al., 1990; Hkansson et al., 2001; Vajdic et al., 2001, 2002) reported statistically significant asso-ciations between a measure of sun exposure and ocular melanoma. Tucker et al. (1985) observed an increased risk of ocular melanoma in people born in the south of the USA (south of 40N) rela-tive to those born in the north (OR, 2.7; 95%CI: 1.35.9), which appeared to be independent of duration of residence in the south. Seddon et al. (1990) reported on two separate series of cases and controls. In the first series, increased risks of uveal melanoma with residence in the south of the USA were observed (OR, 2.4; 95%CI: 1.44.3

44


for up to 5 years; and OR, 2.8; 95%CI: 1.16.9 for more than 5years). In the second series, the risk increased with increasing years of intense sun exposure (OR, 1.5; 95%CI: 1.02.2 for 140 years; and, OR, 2.1; 95%CI: 1.43.2 for > 40 years); this association was only weakly present in the first series; the odds ratio for uveal mela-noma with birthplace in the south of the USA was 0.2 (95%CI: 0.00.7), which was statistically independent of the positive association between duration of residence in the south and uveal melanoma risk. Vajdic et al. (2001, 2002) found that the risk of choroid and ciliary body mela-noma was increased in the highest categories of total sun exposure (OR, 1.6; 95%CI: 1.02.6), weekdays sun exposure (OR, 1.8; 95%CI: 1.12.8), and occupational sun exposure (OR, 1.7; 95%CI: 1.12.8); the underlying trends across quarters of exposure were reasonably consistent and statistically significant. These associations were largely due to stronger associations confined to men. Finally, the one cohort study (Hkansson et al., 2001), based in the Swedish construction industrys health service, observed an increasing risk of ocular melanoma with increasing occupa-tional sun exposure based on recorded job tasks (RR, 1.4; 95%CI: 0.73.0, for medium sun expo-sure; and, RR, 3.4; 95%CI: 1.110.5, for high sun exposure).

Five of the casecontrol studies limited their study to uveal melanoma (melanoma in the choroid, ciliary body, and iris), and one of these excluded iris melanoma because of small numbers. Two studies reported results for iris melanoma (Tucker et al., 1985; Vajdic et al., 2002). One study observed odds ratios of 35 for iris melanoma with the use of an eye shade when outdoors occasionally, rarely or never, relative to almost always (Tucker et al., 1985), and the other observed an increased risk of iris mela-noma in farmers (OR, 3.5; 95%CI: 1.28.9; Vajdic et al., 2002). One study also reported results for conjunctival melanoma, but found no positive

associations with measures of sun exposure (Vajdic et al., 2002).

(c) Meta-analyses

Shah et al. (2005) and Weis et al. (2006) reported the results of meta-analyses of risk of ocular melanoma in relation to sun sensitivity characteristics and sun exposure, including both casecontrol and cohort studies (Table2.10 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.10.pdf). A fixed-effects model was used except when statistically significant heterogeneity was found between the effects of individual studies and a random-effects model was used instead. A summary relative risk was reported only when four or more studies were included in the anal-ysis. In the analysis by Shah et al. (2005), neither latitude of birth nor outside leisure was appre-ciably associated with ocular melanoma. There was weak evidence that occupational exposure to the sun increased ocular melanoma risk (RR for highest exposed category, 1.37; 95%CI: 0.961.96). [The Working group noted that this analysis did not include results of Lutz et al. (2005) or Schmidt-Pokrzywniak et al. (2009), but included those of Gunel et al. (2001), which are a component of Lutz et al. (2005). When the results of Lutz et al. (2005) are substituted for those of Gunel et al. (2001) and those of Schmidt-Pokrzywniak et al. (2009) added to the fixed effects meta-analysis, the meta-RR is 1.25 (95%CI: 1.021.54).]

The meta-analysis of Weis et al. (2006) provides strong evidence that having blue or grey eyes, fair skin and/or burning easily rather than tanning when exposed to the sun are associated with an increased risk of ocular melanoma. Hair colour was not associated with this cancer.

2.1.5 Other sites

Prompted at least in part by the hypotheses arising from ecological studies, casecontrol and cohort studies have been conducted in

45

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.10.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.10.pdf


which measures of personal exposure to solar radiation (loosely referred to here as sun or sunlight exposure) have been related to cancers in internal tissues (Table 2.11 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.11.pdf and Table 2.12 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.12.pdf). Studies that infer high sun exposure from a past history of skin cancer (basal cell carcinoma, squamous cell carcinoma or melanoma) were excluded (see for example, Tuohimaa et al., 2007). It has been argued in respect of these studies that the incidence of second cancers in individuals is elevated by several known and unknown mech-anisms, including common etiological factors and predispositions, and influenced by possible biases in the ascertainment of second cancers [] The net direction of these influences will mostly be in the direction of elevated occurrence of second cancers, against which a possible effect of sunlight and vitamin D [] could be difficult to detect. (IARC, 2008). Thus, such studies are unlikely to be a reliable source of evidence for determining whether sun exposure causes or prevents any other cancers.

(a) Cancer of the colorectum

Two casecontrol studies have related esti-mates of individual sun exposure to risk of cancer of the colorectum. Based solely on death certifi-cates, Freedman et al. (2002) observed a some-what reduced risk (OR, 0.73; 95%CI: 0.710.74) with high ambient sunlight in the state of resi-dence at the time of death, adjusted for age, sex, race, occupational sun exposure (inferred from usual occupation), physical activity, and socio-economic status. In a large population-based study in which participants were interviewed, no appreciable association was found between cancer of the colon and sun exposure recalled for each season for the 2 years before case diag-nosis. With the exception of the second quintile of exposure in women (OR, 1.3), the odds ratios

for each quintile of exposure in each sex varied from 0.91.1, and were not significantly increased (Kampman et al., 2000).

(b) Cancer of the breast

Three casecontrol and two cohort studies have examined the association between meas-ures of sun exposure and breast cancer. In three studies reporting results for sun expo-sure assessed from location of residence, one found slightly higher risks in women residing in California (using south as a reference; Laden et al., 1997); the other two studies found reduced relative risks (0.73 and 0.74) with residence in areas of high mean daily solar radiation (John et al., 1999; Freedman et al., 2002), significantly so in one of these studies (Freedman et al., 2002). Sun-related behaviour was recorded in three studies (John et al., 1999; Freedman et al., 2002; Knight et al., 2007) and was inversely associated with risk for breast cancer for some measures. For example, the relative risks for breast cancer with frequent recreational and occupational sun exposure relative to rare or no exposure were 0.66 (95%CI: 0.440.99) and 0.64 (95%CI: 0.41, 0.98), respectively, in 5009 women from the NHANES Epidemiologic Follow-up Study (John et al., 1999). For the highest category of estimated lifetime number of outdoor activity episodes at 1019 years of age, the odds ratio was 0.65 (95%CI: 0.500.85) in a large Canadian casecontrol study (Knight et al., 2007). In each study, these effect measures were adjusted for a measure of socioeconomic status and some other variables associated with breast cancer.

(c) Cancer of the ovary

In a casecontrol study, based on death certificates, the relative risk of cancer of the ovary was reduced in those residing in areas with high mean daily solar radiation (OR, 0.84; 95%CI: 0.810.88), but not in those with high occupa-tional sun exposure (Freedman et al., 2002).

46

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.11.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.11.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.12.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.12.pdf


(d) Cancer of the prostate

Four casecontrol studies (two hospital-based) and one cohort study (John et al., 2004, 2007) examined the association between meas-ures of sun exposure and risk for cancer of the prostate. In one casecontrol study conducted in two consecutive periods and with patients with benign prostatic hypertrophy as controls, the odds ratio for prostate cancer with highest lifetime sun exposure was [0.32 (95%CI: 0.200.51); combined odds ratio calculated from two reported odds ratios]. Odds ratios were similarly low with indirect measures of sun exposure, such as regular foreign holidays or childhood sunburn (Luscombe et al., 2001; Bodiwala et al., 2003). Two other studies showed weaker evidence of an inverse association of residence in a high solar radiation environment with cancer of the pros-tate (Freedman et al., 2002; John et al., 2004, 2007). Outdoor occupation, self-reported recrea-tional sun exposure, physician-assessed sun exposure or actinic skin damage had no effect on prostate cancer risk in these studies. In a casecontrol study that included only cases of primary advanced cancer of the prostate (John et al., 2005), a reduced risk for cancer of the prostate was reported with high values of sun exposure index (based on comparison of the measured reflect-ance of usually exposed and usually unexposed skin; OR, 0.51; 95%CI: 0.330.80), but with little evidence of similar associations with residential ambient solar radiation or total or occupational lifetime outdoor hours.

(e) Non-Hodgkin lymphoma and other lymphomas

While some early, mainly ecological studies, suggested that sun exposure might increase risk for non-Hodgkin lymphoma, studies of indi-vidual sun exposure suggest that recreational sun exposure may decrease its risk.

Two earlier studies in individuals assessed sunlight exposure based on place of residence,

occupational title and, in one study, industry (Freedman et al., 1997; Adami et al., 1999). The results for residential exposure were conflicting: one study, in the USA, found a reduced relative risk with residence at lower latitudes (Freedman et al., 1997); and the other, in Sweden, an increased risk (Adami et al., 1999). They concurred, however, in finding reduced relative risks in people with high occupational sun exposure with values of 0.88 (95%CI: 0.810.96) in the USA and 0.92 (95%CI: 0.880.97; combined result for men and women) in Sweden. Subsequent studies focusing specifically on occupational sun exposure have not observed a reduced risk of non-Hodgkin lymphoma with higher exposure (van Wijngaarden & Savitz, 2001; Tavani et al., 2006; Karipidis et al., 2007). A study of non-Hodgkin lymphoma in children reported a reduced risk in those who had spent 15 or more days annually at seaside resorts, with an odds ratio of 0.60 (95%CI: 0.430.83; Petridou et al., 2007).

All other studies (Hughes et al., 2004; Smedby et al., 2005; Hartge et al., 2006; Soni et al., 2007; Weihkopf et al., 2007; Zhang et al., 2007; Boffetta et al., 2008; Kricker et al., 2008) were included in a pooled analysis of original data from 8243 cases of non-Hodgkin lymphoma and 9697 controls in ten member studies of the InterLymph Consortium (Kricker et al., 2008; Table2.13 available at http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.13.pdf). [The Working Group noted that results on sun exposure and non-Hodgkin lymphoma in three of these studies have not yet been published separately.] In eight studies in which a composite measure of total sun exposure (recreational plus non-recreational exposure) could be defined, the pooled odds ratio fell weakly with increasing sun exposure to 0.87 (95%CI: 0.711.05) in the fourth quarter of exposure. There was a steeper down-trend for recreational exposure to an odds ratio of 0.76 (95%CI: 0.630.91; P for trend, 0.005), and no appreciable downtrend for non-recreational exposure. Physical activity and obesity, which

47

http://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.13.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.13.pdfhttp://monographs.iarc.fr/ENG/Monographs/vol100D/100D-01-Table2.13.pdf


might be confounding, were not controlled for in the analysis of any of the pooled studies.

Four casecontrol studies have reported on the association between sun exposure and Hodgkin lymphoma (Table 2.11 on-line); there was no consistent pattern of decreasing or increasing risk with different sun exposure measures (Smedby et al., 2005; Petridou et al., 2007; Weihkopf et al., 2007; Grandin et al., 2008). The same was true for multiple myeloma in two casecontrol studies (Boffetta et al., 2008; Grandin et al., 2008). One study found weak evidence of an increased risk of mycosis fungoides [a cutaneous lymphoma] in people with high occupational sun exposure [OR: 1.3 (95%CI: 1.01.9; combined result for men and women] (Morales-Surez-Varela et al., 2006).

2.2 Artificial UV radiation

2.2.1 Use of artificial tanning devices (sunlamps, sunbeds, solaria)

(a) Cutaneous melanoma, squamous cell carcinoma, and basal cell carcinoma

Two meta-analyses of skin cancer in rela-tion to sunbed use have been undertaken over the past few years (Table2.14). The first (IARC, 2006a, 2007a) was based on 19 informative published studies (18 casecontrol, of which nine population-based, and one cohort, all in light-skinned populations) that investigated the association between indoor tanning and skin cancers, and included some 7355 melanoma cases (Table 2.14). The characterization of the exposure was very varied across reports. The meta-relative risk for ever versus never use of indoor tanning facilities from the 19 studies was 1.15 (95%CI: 1.001.31); results were essentially unchanged when the analysis was restricted to the nine population-based casecontrol studies and the cohort study. A doseresponse model was not considered because of the heterogeneity among the categories of duration and frequency

of exposure used in the different studies. All studies that examined age at first exposure found an increased risk for melanoma when exposure started before approximately 30 years of age, with a summary relative risk estimate of 1.75 (95%CI: 1.352.26) (Table2.14). The second meta-analysis (Hirst et al., 2009) included an additional nested casecontrol study of melanoma (Han et al., 2006), bringing the total number of melanoma cases to 7855, and the summary relative risk for melanoma in relation to ever versus never use of sunbeds was reported as 1.22 (95%CI: 1.071.39).

Regarding basal cell carcinoma and squa-mous cell carcinoma, a meta-analysis of the three studies on ever use of indoor tanning facilities versus never use showed an increased risk for squamous cell carcinoma of 2.25 (95%CI: 1.084.70) after adjustment for sun exposure or sun sensitivity (IARC, 2006a, 2007a). One study had information on age at first exposure of indoor tanning facilities and suggested that the risk increased by 20% (OR, 1.2; 95%CI: 0.91.6) with each decade younger at first use. The four studies on basal cell carcinoma did not support an asso-ciation with the use of indoor tanning facilities (IARC, 2006a, 2007a).

(b) Ocular melanoma

Four casecontrol studies have reported explicitly on the association of artificial tanning devices and ocular melanoma (Tucker et al., 1985; Seddon et al., 1990; Vajdic et al., 2004; Schmidt-Pokrzywniak et al., 2009; Table 2.15). Odds ratios for the highest exposure categories in each were: 2.1 (95%CI: 0.317.9) (Tucker et al., 1985); 3.4 (95%CI: 1.110.3) and 2.3 (95%CI: 1.24.3) for the population-based comparison and casesibling comparison, respectively (Seddon et al., 1990); 1.9 (95%CI: 0.84.3) (Vajdic et al., 2004); and 1.3 to 2.1 depending on the control category (Schmidt-Pokrzywniak et al., 2009). The only study to analyse doseresponse found evidence of increasing risk with increasing duration of use (P = 0.04) and, less strongly, estimated

48

Solar and UV radiation49

Table 2.14 Meta-analyses of use of artificial tanning devices and skin cancers

Reference, study location & period

Study description Number of cases and controls

Exposure assessment

Exposure categories

Relative risk (95%CI)

Adjustment for potential confounders

Comments

IARC (2007a) Europe, north America and Australia 1971 to 2001

18 casecontrol studies (10 pop-based) and 1 cohort published in 19812005, with exposure assessment to indoor tanning

Cutaneous melanoma: 7355 cases and 11275 controls from casecontrol studies; cohort: 106379 members BCC-SCC (No. of cases not stated) from 5 casecontrol studies

All studies except two presented estimates for ever versus never

Indoor tanning Melanoma All analyses adjusted for the maximum of potential confounders

One study presented results for men and women separately; Doseresponse was not considered because of the heterogeneity among the categories of duration and frequency of exposure between studies.

Never use 1.0Ever use 1.15 (1.001.31)Age first useNever 1.0

IARC M

ON

OG

RAPH

S 100D

50

Table 2.15 Casecontrol studies of exposure to artificial tanning devices and ocular melanoma

Reference, study location and period

Cases Controls Exposure assessment Exposure categories

Relative risk (95% CI)

Adjustment for potential cofounders

Comments

Tucker et al. (1985), USA, 197479

439 White patients with intraocular melanoma confirmed histologically or from highly reliable ancillary studies; participation rate, 89%

419 White patients with detached retina not due to tumours; matched by age, sex, race, date of diagnosis; participation rate, 85%

Telephone interview with detailed information about medical history, family history, employment, exposure to environmental agents, sunlight; details from ophthalmologic examination and medical history abstracted from medical records; interview with next-of-kin for 17% of cases and 14% of controls, half of them with spouses

Sunlamp use Age, eye colour and history of cataract

Never 1.0Rarely 1.3 (0.82.3)Occasionally 1.3 (0.53.6)Frequently 2.1 (0.317.9)

Seddon et al. (1990), Massachusetts, USA, 198487

White patients with clinically or histologically confirmed melanoma of the choroid, ciliary body or both, identified at local hospital or by mailing to ophthalmologists, diagnosed within previous yr; age range, 1788 yr, mean, 57 yr; participation rate, 89% (see comments)

Series 1: selected by random digit dialing, matched 2:1 by sex, age, city of residence, 85% response rate Series 2: living sibling of cases, up to 4 siblings per case, median, 2; participation rate, 97%

Telephone interview including constitutional factors, ocular and medical histories, and exposure to environmental factors including natural and artificial sources of UV

Casecontrol series 1*

Age, eye and skin colour, moles, ancestry, eye protection, outside work, fluorescent lighting, southern residence, yr of intense exposure

*Series 1: population-based, 197 cases and 385 controls; Series 2: not population-based, 337 cases and 800 sibling controls. 140 cases were included in both series.

Sunlamp useNever 1.0Rarely 0.7 (0.41.4)Occasionally or frequently

3.4 (1.110.3)

Casecontrol series 2*Sunlamp useNever 1.0Rarely 0.9 (0.61.4)Occasionally or frequently

2.3 (1.24.3)


Table 2.15 (continued)



Relative risk (95% CI)

Adjustment for potential cofounders

Comments

Vajdic et al. (2004), Australia, 199698

246 White Australian residents, aged 1879 yr, with histopathologically or clinically diagnosed melanoma originating in the choroid, ciliary body; participation rate, 87% among those eligible

893 controls matched 3:1 by age, sex, residence, selected from electoral rolls; participation rate, 47%

Self-administered questionnaire, and telephone interview regarding sun exposure, sun-protective wear and quantitative exposure to welding equipment and sunlamps

Sunlamp use* Age, sex, place of birth, eye colour, ability to tan, squinting as a child and total personal sun exposure at 10, 20, 30 and 40 yr of age

*Sunlamps use includes use of sunbeds and tanning booths

Never 1.0Ever 1.7 (1.02.8)Duration of use1 mo 1.2 (0.52.8)2mo to 1yr 1.8 (0.83.9)>1yr 2.3 (0.95.6)Lifetime hours of use0.11.4 1.3 (0.53.2)1.57.8 1.8 (0.84.2)>7.8 1.9 (0.84.3)Period of first use1990 4.3 (0.727.9)Age at first use>20 yr 1.5 (0.82.6)20 yr 2.4 (1.06.1)

Schmidt-Pokrzywniak et al. (2009), Germany, 200205

459 cases of incident primary uveal melanoma diagnosed at 1 clinic, aged 2074 yr

Control 1: 827 population-based, selected from mandatory list of residence, matched 2:1 on age (5-yr age groups), sex and region Control 2: 187 sibling controls, matched 1:1 by (+/ 10 yr) and sex when possible

Self-administered postal questionnaire and computer-assisted telephone interview

Regular sunlamp use

Results presented for population controls. Odds ratios with sibling controls were somewhat higher, but with wider confidence intervals and not significant; *Sunlamps use includes use of sunbeds and tanning booths

No 1.0 Yes 1.3 (0.91.8)Age at first useNever used 1.0>20 yr 1.3 (0.91.9)


cumulative time of exposure (P=0.06) (Vajdic et al., 2004). The two most recent studies (Vajdic et al., 2004; Schmidt-Pokrzywniak et al., 2009) calculated odds ratios for exposure that started at or before 20 years of age and after this age; in both, the odds ratio was greater for exposure starting at the younger age. The results of Seddon et al. (1990) and Vajdic et al. (2004) were adjusted for sun sensitivity and personal sun exposure. [The Working Group noted that Schmidt-Pokrzywniak et al. (2009) found little evidence of associations between measures of personal sun exposure and ocular melanoma.]

(c) Internal cancers

Five casecontrol studies (Table 2.16) have reported on the association of the use of artifi-cial tanning devices and cancer of the breast (one study), non-Hodgkin lymphoma (four studies), Hodgkin lymphoma (three studies), multiple myeloma (two studies), and lymphoprolifera-tive syndrome (one study) (Smedby et al., 2005; Hartge et al., 2006; Knight et al., 2007; Boffetta et al., 2008; Grandin et al., 2008). In all the studies of non-Hodgkin lymphoma, the risk was lower in people who had used artificial tanning devices than in those who had not; in two there was also a doseresponse relationship across exposure categories with a P value for trend of 0.01 (Smedby et al., 2005; Boffetta et al., 2008). Odds ratios were also below unity for cancer of the breast (Knight et al., 2007) and for Hodgkin lymphoma (Smedby et al., 2005; Boffetta et al., 2008), with a significant doseresponse relation-ship (P value for trend = 0.004) in one study of Hodgkin lymphoma (Smedby et al., 2005). Confounding with exposure to natural sunlight cannot be ruled out as an explanation for these inverse relationships because none of the studies adjusted the results for sun exposure.

2.2.2 Welding

Six separate casecontrol studies (seven reports) and one meta-analysis have reported on associations between welding and risk of ocular melanoma (Table 2.17). All studies reported an odds ratio for ocular melanoma above unity in most categories of exposure to welding. Seddon et al. (1990) reported on two sets of cases and controls and found an increased risk in only one of them. Lutz et al. (2005) found an increased risk with a history of at least 6months employ-ment in welding or sheet metal work, but not for working with welding; the increase observed was restricted to the French component of the study, which Gunel et al. (2001) had previously reported. The strongest associations of welding with ocular melanoma (although based on small numbers) were reported in those studies that restricted the exposure definition to work as a welder, i.e. not including being in proximity to welding (Tucker et al., 1985; Siemiatycki, 1991; Gunel et al., 2001; Lutz et al., 2005). Several studies showed evidence of doseresponse rela-tionships (Holly et al., 1996; Gunel et al., 2001; Vajdic et al., 2004) with duration of employment or of use.

The meta-analysis (Shah et al., 2005) estimated a meta-relative risk of 2.05 (95%CI: 1.203.51) for welding, using a random-effects model. [The Working Group noted that this study included results from Ajani et al. (1992), which overlap with those from casecontrol Series 1 of Seddon et al. (1990), and did not include those from the casecontrol Series 2 of Seddon et al. (1990). It also did not include results from Siemiatycki (1991).]

2.3 UVA, UVB, and UVC

Epidemiology has little capacity to distinguish between the carcinogenic effects of UVA, UVB, and UVC. UVC is not present in natural sunlight at the surface of the earth and is therefore not

52


relevant; in almost all circumstances humans are exposed simultaneously to UVB and UVA, and UVB and UVA exposures vary more or less in parallel (see Section 1). Several epidemiological approaches have been used in an attempt to distinguish the effects of UVA and UVB on skin cancer risk. Their major focus has been to assess whether solar UVA exposure contributes to the increased risk of cutaneous melanoma, for which there is some conflicting evidence in experimental studies (see Section 4). These include studies on exposure to UVA for artificial tanning, effect of sunscreens on melanoma risk, and UVB photo-therapy without associated exposure to PUVA (psoralen-UVA photochemotherapy).

PUVA is the combination of psoralen with UVA radiation, and is used in the treatment of psoriasis. PUVA has been reviewed previously by two IARC Working Groups and there is sufficient evidence that PUVA therapy is carcinogenic to humans (Group 1), causing cutaneous squamous cell carcinoma (IARC, 1986, 2012), and these studies will not be reviewed here.

2.3.1 Descriptive studies

Garland et al. (1993) noted that rising trends in the incidence of and mortality from melanoma have continued since the 1970s and 1980s, when sunscreens with high sun protec-tion factors became widely used. They related this observation to the fact that commonly used chemical sunscreens had blocked UVB but not UVA; and the possibility that by preventing erythema, sunscreens would permit extended sun exposure and thus substantially increase exposure to UVA. However, nearly half of the melanoma mortality increase between 195054 and 199094 in the USA in white men and more than half of that in white women had occurred by 197074, with only a minor upward pertur-bation in the trend after 197074. Thus, there probably was not a close association between

increasing use of sunscreens blocking UVB and the increasing risk of melanoma.

Moan et al. (1999) plotted the relationships of UVB and UVA irradiances and incidence rates of cutaneous basal cell carcinoma, squamous cell carcinoma and melanoma using data from Australia, Canada, the Czech Republic, Denmark, Finland, Iceland, Norway, New Zealand, Sweden, Scotland, USA, and the United Kingdom. As expected, all were inversely related to latitude but the slope of the fitted linear relationship was numerically smaller for UVA than for UVB, and for melanoma than for basal cell carcinoma and squamous cell carcinoma. Estimates of biological amplification factors (relative increase in risk per unit increase in exposure) based on these slopes for UVB were, in men and women respectively, 2.8 and 2.8 for basal cell carcinoma, 3.1 and 2.9 for squamous cell carcinoma, and 1.3 and 1.0 for melanoma. Those for UVA and melanoma were 3.8 and 2.9, respectively, suggesting that UVA may play a significant role in the induction of melanomas.

2.3.2 Exposure to artificial UVA for tanning purposes

Early artificial tanning devices emitted both UVB and UVA. UVB emissions were subse-quently reduced relative to UVA, presumably to reduce skin cancer risk, but have been increased again recently to mimick the sun and to produce longer lasting tans (see Section 1). In principle these periods of different relative exposures to UVA and UVB during artificial tanning could be used to evaluate the relative effects of UVA and UVB on skin cancer risk. Veierd et al. (2003, 2004) attempted this analysis in a cohort study of Norwegian and Swedish women who had reported their use of a sunbed or sunlamp (solarium) in different age periods on entry to the cohort. They defined three subgroups of women: those who had used solaria in the period 196383 (mainly before they became mainly

53

IARC M

ON

OG

RAPH

S 100D

54

Table 2.16 Associations of use of artificial tanning devices with cancers of internal tissuesa


Cancer type Exposure assessment

Exposure categories Relative risk

Knight et al. (2007), Canada, 200304

Breast cancer Telephone interview Ever sunlamp useAge 1019No 1.0Yes 0.81 (0.571.14)Age 2029No 1.0Yes 0.88 (0.661.18)Age 4554No 1.0Yes 0.84 (0.641.11)

Hartge et al. (2006), USA, 19982000

Non-Hodgkin lymphoma

Self-administered questionnaire and computer assisted personal interview

Use of sunlamp or tanning boothNever 1.0Ever 0.88 (0.661.19)Only after age 20 0.97 (0.691.37)Before age 20 0.72 (0.451.14)



Cancer type Exposure assessment

Exposure categories Relative risk

Boffetta et al. (2008), France, Germany, Ireland, Italy, and Spain, 19982004


Interviewer administered questionnaire

Sunlamp useNever 1.0124 times 0.79 (0.591.04)25 times or more 0.69 (0.510.93)

Hodgkin lymphoma Never 1.0124 times 0.86 (0.531.39)25 times or more 0.93 (0.571.50)

Multiple myeloma Never 1.0

124 times 0.76 (0.411.41)25 times or more 1.10 (0.592.05)

Grandin et al. (2008), France, 200004


Self and interviewer administered questionnaires

Aesthetic use of artificial UV radiationNo 1.0Yes 1.1 (0.71.7)Regularly 0.5 (0.21.3)Occasionally 1.4 (0.82.3)

Hodgkin lymphoma No 1.0Yes 1.6 (0.73.6)Regularly 0.6 (0.13.3)Occasionally 2.2 (0.95.5)

Lymphoproliferative syndrome

No 1.0Yes 1.5 (0.73.5)Regularly 0.9 (0.24.6)Occasionally 1.9 (0.74.7)

Multiple myeloma No 1.0Yes 1.2 (0.43.6)Regularly 0.8 (0.17.3)Occasionally 1.4 (0.44.9)

a In none of these studies was potential confounding with exposure to natural sunlight controlled in the analysisyr, year or years


IARC M

ON

OG

RAPH

S 100D

56

Table 2.17 Casecontrol studies on welding and ocular melanoma



Relative risk Adjustment for potential confounders

Comments

Tucker et al. (1985), USA, 197479

439 White patients with intraocular melanoma confirmed histologically or from highly reliable ancillary studies; participation rate, 89%

419 White patients with detached retina not due to tumours; matched by age, sex, race, date of diagnosis; participation rate, 85%

Telephone interview with detailed information about medical history, family history, employment, exposure to environmental agents, sunlight; details from ophthalmologic examination and medical history abstracted from medical records; interview with next-of-kin for 17% of cases and 14% of controls, half of them with spouses

Ever worked as a welder

Age, eye colour and history of cataractNo 1.0

Yes 10.9 (2.156.5)

Seddon et al. (1990), Massachusetts, USA, 198487

White patients with clinically or histologically confirmed melanoma of the choroid, ciliary body or both, identified at local hospital or by mailing to ophthalmologists, diagnosed within previous yr; age range, 1788 yr, mean, 57 yr; participation rate, 89% (see comments)

Series 1: selected by random digit dialing, matched 2:1 by sex, age, city of residence, 85% response rate Series 2: living sibling of cases, up to 4 siblings per case, median, 2; participation rate, 97%

Telephone interview including constitutional factors, ocular and medical histories, and exposure to environmental factors including natural and artificial sources of UV

Casecontrol series 1

Age, eye and skin colour, moles, ancestry, use of sunlamps, eye protection, outside work, fluorescent lighting, southern residence, yr of intense exposure

Series 1: population-based, 197 cases and 385 controls Series 2: not population-based, 337 cases and 800 sibling controls. 140 cases were included in both series. Result for case series 1 also was reported by Ajani et al. (1992) using the same numbers but with fewer covariates in the logistic regression model (see below).

Exposure to welding arcNo 1.0Yes 1.3 (0.53.1)Casecontrol series 2Exposure to welding arcNo 1.0Yes 0.9 (0.61.5)





Comments

Siemiatycki (1991), Montreal, Canada, 197985

[33] incident male cases of uveal melanoma, aged 3570 yr, histologically confirmed; response rate, 69.6%

533 population controls; participation rate, 72%

Personal interview and collection of detailed occupational history

Occupational exposure to arc welding fumes

Age, family income, ethnicity, respondent type, cigarette and alcohol indexes

4 exposed cases

No 1.0Yes 8.3 (2.527.1)

Ajani et al. (1992), USA, 198487

197 White patients with uveal melanoma, histologically confirmed, diagnosed during the previous yr, residents of 6 New England States; mean age, 59.2 yr, range 1888 yr; participation rate, 92%

385 controls selected by random digit dialling, matched 2:1 for age (+/ 8 yr), sex, telephone exchange; mean age, 58.3 yr, range 1988 yr; response rate, 85%

Telephone interview with occupational history and exposures related to work occurring 15 yr before the interview.

Exposure to welding arc

Age, ancestry, skin colour, moles, use of sunlamps, past income level

Same population as in study by Seddon et al. (1990) in case series 1 using the same numbers but with more covariates in the logistic regression model (see above).

No 1.0Yes 0.99 (0.48

2.05)

Holly et al. (1996), USA, 197887

221 male White patients with histologically confirmed uveal melanoma, age 2074 yr residing in 11 States; participation rate, 93%

447 controls selected by random digit dialling, matched 2:1 by age (5-yr age group) and residential area; interview rate, 77%

Interviewer administered questionnaire with demographic and phenotypic caracteristics, occupational history, exposure to chemicals.

Welding* Age, number of large nevi, eye colour, tanning or burning response to 30 min. sun exposure in the summer noond sun

* Self welding or in proximity to others for >3h a wk for >6mo

No 1.0Yes 2.2 (1.33.5)Years from start of occupation to diagnosis or interview10 1.2 (0.26.6)1129 1.5 (0.73.0)30 2.1 (1.14.0)


IARC M

ON

OG

RAPH

S 100D

58




Comments

Gunel et al. (2001), France 199596

50 cases (29 men and 21 women) identified from records of local pathology departments for surgery, and from 2 cancer treatment centres in France; diagnosis confirmed by pathologists or ophthalomogic report; participation rate, 100%

479 (321 men, 158 women) controls selected from electoral rolls, frequency matched by age (5-yr interval), sex and study area; participation rate, 76%

Face-to-face interview, or occasionally telephone interview

Worked for six mo or more as a welder or sheet metal worker

Age Data also included in analysis of Lutz et al. (2005). Results shown here for men only; only one woman in this study had worked as a welder and she was a case.

No 1.0Yes 7.3 (2.620.1)Duration of employment as a welderLess than 20 yr 5.7 (1.619.8)20 yr or more 11.5 (2.455.5)

Vajdic et al. (2004), Australia, 199698

246 White Australian residents, aged 1879 yr, with histopathologically or clinically diagnosed melanoma originating in the choroid, ciliary body; participation rate, 87% among those eligible

893 controls matched 3:1 by age, sex, residence, selected from electoral rolls; participation rate, 47%

Self-administered questionnaire, and telephone interview regarding sun exposure, sun-protective wear and quantitative exposure to welding equipment and sunlamps

Own welding Age, sex, place of birth, eye colour, ability to tan, squinting as a child and total personal sun exposure at 10, 20, 30, and 40 yr of age

Never 1.0Ever 1.2 (0.81.7)Duration of use0.14.0 yr 0.8 (0.41.4)4.1 to 22.0yr 1.2 (0.72.2)>22 yr 1.7 (1.02.7)Lifetime hours of use0.152.0 1.1 (0.61.9)52.1858.0 1.4 (0.82.3)>858 1.1 (0.61.9)Age at first use>20 yr 1.2 (0.81.9)20 yr 1.2 (0.71.9)






Comments

Lutz et al. (2005), Denmark, Latvia, France, Germany, Italy, Sweden, Portugal, Spain, and the United Kingdom, 199596

292 incident cases of uveal melanoma, identified from ophthalmologic departments, hospital records or cancer registries aged 3569 yr; participation rate, 91%

2062 population controls selected from population registers, electoral rolls or practitioner, frequency matched by region, sex and 5-yr birth cohorts; participation rate, 61%; 1094 cancer controls randomly selected from colon cancer patients; participation rate, 86%

Questionnaire with face-to-face or telephone interview

Worked for six mo or more as a welder or sheet metal worker

Data from France reported in analysis of Gunel et al. (2001). Results shown here for men only; only one woman in this study had worked as a welder and was a case.

No 1.0Yes 2.2 (1.24.0)Working with weldingNo 1.0Yes 0.9 (0.61.5)

d, day or days; h, hour or hours; min, minute or minutes; mo, month or months; wk, week or weeks; yr, year or years



UVA-emitting), the period 197991 (mainly after solaria were designed to emit mainly UVA) or the period 197587 (covering both catego-ries of solarium) when they were 2029 years of age. The odds ratios for solarium use in these subgroups were 3.75 (95%CI: 1.738.13) for use in 196383, 3.19 (95%CI: 1.228.32) for use in 197991, and 1.28 (95%CI: 0.463.60) for use in 197587. These results show little difference between those exposed in the earlier and later periods of solarium use. [The Working Group noted that only seven cases of melanoma were observed in each of these periods, and there was little statistical power to see a difference.] A recent meta-analysis of use of artificial tanning devices and skin cancer (IARC, 2007b) reported that the relative risks of melanoma associated with ever use of a sunbed or sunlamp did not vary with year of publication of a study or the first year of a study period, where available. [The Working Group noted that the most relevant time metric would be year of first reported use of a sunbed or sunlamp, rather than the year of publication or first year of study period.]

2.3.3 Use of sunscreens and risk for melanoma

Initially, sunscreens contained only UVB absorbers; more recently they have covered a broader spectrum with the addition of UVA reflectors or absorbers, although many are still less effective against the higher wavelengths of UVA than they are against UVB (see Section 1). Recent meta-analyses of published observa-tional studies of sunscreen and melanoma, each including slightly different subsets of studies, have found meta-relative risks close to unity with highly significant heterogeneity among studies: 1.11 (95%CI: 0.373.32) with a P value for hetero-geneity


patients who had less than 100 PUVA treatments, the incidence rate ratio for cutaneous squamous cell carcinoma with 300 UVB treatments was 0.81 (95%CI: 0.341.93) for chronically sun-exposed sites, and 2.75 (95%CI: 1.116.84) for rarely to intermittently sun-exposed sites. The corresponding values for basal cell carcinoma were 1.38 (95%CI: 0.802.39) for chronically sun-exposed sites and 3.00 (95%CI: 1.306.91) for intermittently sun-exposed sites. [The Working Group noted that the possibility that the observed effect required interaction with PUVA or another treatment for psoriasis cannot be ruled out in this study.] Hearn et al. (2008) described the results of follow-up of 3867 patients who had received narrow-band UVB phototherapy, a quarter of whom had also received PUVA. In comparison with data from the Scottish Cancer Registry, there were near 2-fold increases in the risk of first squamous cell carcinoma [two observed cases] and of first basal cell carcinoma [14 observed cases] for treatment with narrow-band UVB only, but their 95% confidence intervals included unity. For melanoma, the relative risk was just below 1. For those who had more than 100 UVB therapy treatments, the risks, relative to those who received 25 or less such treatments, were 1.22 (95%CI: 0.284.25) for basal cell carcinoma, 2.04 (95%CI: 0.1717.8) for squamous cell carcinoma, and 1.02 (95%CI: 0.0212.7) for melanoma. Two previous small studies of narrow-band UVB, of 126 (Weischer et al., 2004) and 484 patients (Black & Gavin, 2006), observed only one skin cancer between them, an in-situ melanoma, in less than 10 years of follow-up.

Given the few cases of skin cancer so far reported in patients given UVB phototherapy as their only form of phototherapy, the statistical power of currently available studies to detect other than a large increase in relative risk of any type of skin cancer with this therapy, and, there-fore, of UVB specifically is weak.

2.4 Synthesis

2.4.1 Solar radiation

In Caucasian populations, both basal cell carcinoma and squamous cell carcinoma are strongly associated with solar radiation, as meas-ured by indicators of accumulated solar skin damage (e.g. increasing age, especially for squa-mous cell carcinoma; and presence of actinic keratoses), and secondarily by recalled episodes of acute solar skin damage (multiple sunburns).

The causal association of cutaneous mela-noma and solar exposure is established, this link has become clearer in the last decade or so through the observation of the site-specific heterogeneity of melanoma, the lower-than-average phenotypic risk for skin carcinogenesis among outdoor workers, and the recognition that the different associations of melanoma with sun exposure observed among Caucasian people at different latitudes around the world correlate with marked variations in sun exposure oppor-tunity and behaviour.

Five casecontrol studies of cancer of the lip have been published. The three earliest studies found apparent increases in risk with outdoor work, but use of tobacco could not be ruled out as an explanation for these associa-tions. The two later studies both took account of possible confounding of outdoor exposure with tobacco smoke. One of them, in women, showed increased risks for cancer of the lip with several measures of exposure, together with strong and moderately consistent doseresponse relation-ships. The other, in men, found no increase in risk with leisure time or holiday sun exposure but a substantial increase in risk with cumula-tive exposure during outdoor work during the summer months, without any indication of doseresponse across four categories. This lack of doseresponse suggests bias rather than a causal effect.

61


Four casecontrol studies reported at least one result each suggesting that sun exposure is associated with conjunctival intraepithelial neoplasia or squamous cell carcinoma of the eye. Only one study was exclusively of conjunc-tival squamous cell carcinoma; in this study and another, the relevant exposure variables (office work and cultivating the fields) were only indirect measures of sun exposure. A very large difference between cases and controls in preva-lence of conjunctival solar elastosis in another study raised concerns about possible bias. The remaining study reported a strong association of ocular surface dysplasia with solar keratoses and increasing risk with increasing duration of residence at 30 south latitude. However, only 22% of its cases had conjunctival squamous cell carcinoma.

Two out of three studies that examined the distribution of choroidal melanomas found them to be concentrated in the central area or the macula area of the choroid, which coincides with the estimated distribution of light in the retinal sphere. Of ten casecontrol studies of ocular melanoma published from 1985 to 2009, four reported statistically significant associa-tions of one or more measures of sun exposure with ocular melanoma. In two studies, these associations were with the latitude of birth or of residence in early life, with some inconsistency between them. In the other two, which were more recent and had better measures of exposure than many previous studies, one study related only to occupational sun exposure and showed a strong association with a doseresponse relationship, and the strongest association seen in the other was with occupational sun exposure and showed evidence of a doseresponse relationship. These results relate principally to choroid and ciliary body melanomas (the dominant types). Two studies reported results consistent with a posi-tive association of small numbers of iris mela-nomas with sun exposure. One study with a

small number of conjunctival melanomas found no such association.

The associations of sun exposure with several internal cancers have been investigated in casecontrol and cohort studies, generally with the hypothesis that sun exposure might be protective against such cancers. The cancers investigated included cancer of the colorectum (two studies), of the breast (five studies), of the ovary (one study), of the prostate (four studies), and several cancers of the lymphatic tissue, principally non-Hodgkin lymphoma and Hodgkin disease (15 studies). Exposure metrics used in these studies included residential or occupational ambient solar radiation, recreational or non-recreational sun exposure, recent and lifetime sun exposure, and sun-related behaviour. The results were mostly inconsistent.

2.4.2 Artificial sources of UV

(a) Tanning appliances

Two meta-analyses investigated the associa-tion between indoor tanning and skin cancers.

The summary relative risk for ever versus never use of indoor tanning facilities was significantly increased for melanoma, with no consistent evidence for a doseresponse relation-ship. All studies that examined age at first expo-sure found an increased risk for melanoma when exposure started before approximately 30 years of age, with a summary relative risk estimate of 1.75.

For squamous cell carcinoma, the three available studies found some evidence for an increased risk, especially when age at first use was below 20 years. Studies on basal cell carci-noma did not support an association with use of indoor tanning facilities.

Four casecontrol studies reported on asso-ciations between artificial tanning devices and ocular melanoma. Each observed an increase in risk of ocular melanoma in the highest category of exposure to these devices, and there were

62


indications of a doseresponse relationship in three of the studies. In two studies, the risk was higher in people who began exposure before 20 years of age than those who began after this age. Possible confounding with natural sun exposure was explicitly addressed in two of the studies.

Five studies reported on the association of use of indoor tanning devices with internal cancers, specifically breast cancer, non-Hodgkin lymphoma, Hodgkin lymphoma, and multiple myeloma. Most studies found little evidence of an association. Two studies observed inverse associations between the use of internal tanning devices and non-Hodgkin lymphoma, and one study with Hodgkin lymphoma. Possible confounding with exposure to natural sunlight cannot be ruled out in any of these studies.

(b) Welding

Six casecontrol studies reported on the asso-ciation between welding and ocular melanoma. All found evidence of a positive association, which was strong in three studies, each of which related specifically to working as a welder or sheet metal worker (other studies included working in proximity to welding in the definition of expo-sure). In each of three studies in which it was examined, there was evidence of a doseresponse relationship.

2.4.3 UVA, UVB, UVC

Several sources of evidence were examined to see if the carcinogenic effects of UVA and UVB could be distinguished: descriptive studies of skin cancer have shown that the slope of lati-tude variation in incidence of melanoma is less than that in incidence of squamous cell carci-noma and basal cell carcinoma, suggesting that melanoma incidence is more influenced by UVA irradiance than are squamous cell carcinoma and basal cell carcinoma. Present data on the risk for melanoma associated with the of UV-emitting tanning devices show little evidence that it varies

with the relative contributions of UVB and UVA emitted from the devices. There is little or no evidence to suggest that the use of sunscreens that block mainly UVB radiation increased the risk for melanoma. Studies of patients exposed exclusively to UVB phototherapy show weak evidence of an increase in risk of squamous cell carcinoma and basal cell carcinoma, based on a few cases.

3. Cancer in Experimental Animals

The previous IARC Monograph on solar and ultraviolet radiation c

solar and ultraviolet radiation - …monographs.iarc.fr/eng/monographs/vol100d/mono100d-6.pdfsolar...

Documents